PRNG API

random_number.md

Basic random number generators

Language	Basic function	Kind	Range	Open/Close
Fortran	random_number()	real32/64	[0,1)	N/A
C	rand()	integer	[0,RAND_MAX]	N/A
C++	uniform_real_distribution()	real32/64	[a,b)	N/A
Rust	.gen()	real64	[0,1)	N/A
Go	.Float64() and .Float32()	real64	[0,1)	N/A
.Net framework	NextDouble()	real64	[0,1)	N/A
gawk	rand()	real	[0,1)	N/A
Perl	rand()	real	[0,1)	N/A
Ruby	.rand()	real	[0,1)	N/A
Python	random()	real	[0,1)	N/A
NumPy	random()	real64	[0,1)	N/A
Julia	rand()	real64	[0,1)	N/A
MATLAB	rand()	real64	(0,1)	N/A
GNU Octave	rand()	real64	(0,1)	N/A
Scilab	rand()	real64	[0,1)	N/A
R	runif()	real64	(0,1) comment	N/A
SAS	RAND('UNIFORM')	real	(0,1)	N/A
Mathematica	RandomReal[]	real	[0,1] comment	N/A
Maxima	random()	real	[0,1)	N/A
Maple	rand()	real	[0,1]	N/A
Microsoft Excel	RAND()	real	[0,1)	N/A
MKL	vdrnguniform()	real32/64	[0,1)	N/A
NAG	G05SAF()	real32/64	(0,1]	[a,b] by another subroutine
IMSL	RNUN()	real32/64	(0,1)	[0,1) by Mersenne Twister
HSL	FA04AD()	real64	(0,1)	N/A
RANLIB	ranf()	real32	(0,1)	N/A
GSL	gsl_rng_uniform()	real64	[0,1)	(0,1) by another subroutine
Mersenne Twister	genrand_res53()	real64	[0,1)	[0,1] and (0,1)
cuRAND	curand_uniform()	real32/64	(0,1]	N/A
AMD RNG	DRANDUNIFORM()	real64	(0,1] old doc	N/A

API for pseudorandom number generators

C++

• C++11

   // random bit engine with seed
   std::default_random_engine rng(123456789);
   std::cout << "Random bits: " << rng() << '\n';

   // specific engine ex. Mersenne Twister
   // std::mt19937 rng(123456789);
   // use an array for seed
   // std::seed_seq seedarray{1,2,3,4,5};
   // std::mt19937 rng(seedarray);

   // re-seed
   rng.seed(1);

   // distribution with fixed parameters: [0,1)
   std::uniform_real_distribution<double> unif_rand(0, 1);
   // call it with the engine object
   std::cout << "Random numbers [0,1): " << unif_rand(rng) << '\n';

   // changing the parameters [7,10)
   unif_rand.param( decltype(unif_rand)::param_type{ 7, 10 } ) ;
   std::cout << "Random numbers [7,10): " << unif_rand(rng) << '\n';

Rust

• rand::distributions

Similar to C++ (no code here).

Julia

• Julia 1.0 Random
• Julia 1.0 Distributions

using Random
using Distributions

# make a generator
rng = MersenneTwister(123456789);

# random bits
rand(rng,UInt64)

# re-seed
Random.seed!(rng,543210)

# uniform real distribution [0,1)
rand(rng)

# uniform real distribution [7,10)
rand(rng,Uniform(0,10))

NumPy

• NumPy 1.17

from numpy.random import Generator, PCG64, MT19937, SeedSequence

# make a generator
seed = SeedSequence(1234)
rg = Generator(MT19937(seed))
# simple version
# rg = Generator(MT19937(1234))

# random bits
rg.bytes(8)

# re-seed
rg = Generator(MT19937(seed))

# uniform real [0,1)
rg.random()

# uniform real [7,10)
rg.uniform(7,10)

MKL

type(VSL_STREAM_STATE) :: stream
integer :: status,brng,seed,method
integer(int64) :: b
real(real64) :: d

! brng = engine
brng = VSL_BRNG_MT19937
seed = 123456789
status = vslnewstream( stream, brng, seed )

! random bits
method = VSL_RNG_METHOD_UNIFORMBITS64_STD
status = virnguniformbits64( method, stream, 1, b )

! re-seed
status = vslnewstream( stream, brng, seed )

! uniform real [0,1)
method = VSL_RNG_METHOD_UNIFORM_STD
status = vdrnguniform( method, stream, 1, d, 0d0, 1d0 )

! uniform real [7,10)
method = VSL_RNG_METHOD_UNIFORM_STD
status = vdrnguniform( method, stream, 1, d, 7d0, 10d0 )

NAG

integer,parameter :: LSEED=1, LSTATE=17
integer :: GENID, SUBID, SEED(LSEED),STATE(LSTATE), IFAIL
real(real64) :: d

! GENID = engine
GENID = 1  ! default
SEED = 123456789
call G05KFF(GENID, SUBID, SEED, LSEED, STATE, LSTATE, IFAIL)

! no random bits

! uniform real [0,1)
G05SAF(1, STATE, d, IFAIL)

! uniform real [7,10)
call G05SQF(1, 7d0, 10d0, STATE, d, IFAIL)